基于线性回归及BP神经网络的RAT最大释放冲击载荷预测研究  

Research on Predicting the Maximum Deployment Impact Load of RAT Based on Linear Regression and BP Neural Network

作　　者：洪烨 王帮亭[1] 王志伟 杨溢炜 马莹 郦江 HONG Ye;WANG Bang-ting;WANG Zhi-wei;YANG Yi-wei;MA Ying;LI Jiang(Shanghai Aircraft Design and Research Institute,Shanghai,201000,China)

机构地区：[1]上海飞机设计研究院,上海201000

出　　处：《航空计算技术》2024年第5期53-57,共5页Aeronautical Computing Technique

摘　　要：冲压空气涡轮(Ram Air Turbine,RAT)最大释放冲击载荷是飞机结构设计重要参数。当前RAT释放冲击载荷的试验仅试飞或高速风洞试验可以得到,寻找一种有效的RAT最大冲击载荷预测方法很有必要。通过分析得到RAT最大释放冲击载荷的影响因素与飞行高度和飞行空速有直接关系,采用线性回归及BP神经网络研究飞行高度和空速对RAT最大释放载荷的影响,并从平均绝对误差及均方根百分误差进行评价。研究将试验及仿真结果作为训练样本,训练完成后将已知输入层参数输入后预测RAT最大释放载荷。对比预测最大释放载荷与试验结果,线性回归预测值平均绝对误差及均方根百分误差小于10%,BP神经网络预测值平均绝对误差及均方根百分误差的平均值小于5%。The maximum deploy impact load of Ram Air Turbine(RAT)is an important parameter in aircraft structural design.It is necessary to find an effective method for predicting the maximum impact load of RAT,as only flight tests or high speed wind tunnel tests can be conducted to deploy impact load.This article analyzes the factors that affect the maximum deploy impact load of RAT,which are directly related to flight altitude and airspeed.Linear regression and BP neural network are used to study the influence of flight altitude and airspeed on the maximum deploy load of RAT,and the evaluation is conducted from the average absolute error and root mean square percentage error.This study uses experimental and simulation results as training samples,and after training,the known input layer parameters are inputted to predict the maximum deploy load of RAT.Comparing the predicted maximum deploy load with the experimental results,the average absolute error and root mean square percentage error of linear regression prediction values are less than 10%,and the average absolute error and root mean square percentage error of BP neural network prediction values are less than 5%.

关 键 词：冲压空气涡轮 BP神经网络 线性回归 释放冲击载荷 

分 类 号：TM623.8[电气工程—电力系统及自动化]